Samsung has injected capital into Dutch artificial intelligence chip developer Euclyd, taking part in a financing round valued at 231 million US dollars, a move that comes as investors aggressively pour money into alternatives to Nvidia's graphics processing units.
Euclyd chief executive Bernardo Kastrup revealed in an interview on Tuesday that the company's Series A fundraising, which brought in 200 million euros, received backing from Somerset Capital Partners, EQT's European growth fund, and the Innovation Industries Fund, with these institutions co-leading the round alongside Samsung.
The startup, established in 2024, is developing an AI chip system whose architecture diverges from GPU designs, targeting inference scenarios and encompassing both processor and memory architecture. Nvidia's graphics processing units, originally developed for gaming before being repurposed for AI model training and model inference, catapulted the company to become the world's most valuable corporation.
The company currently holds a near-monopoly position in the premium chip market. However, several major cloud providers and startups are now working on custom processors tailored to AI workloads. OpenAI announced in August that its first AI chip, named Jalapeño, delivers "industry-leading speed and energy efficiency." Google, Amazon Web Services, and Meta have also been developing their own AI chips.
Euclyd's chief executive stated: "Artificial intelligence is becoming a cornerstone of economic growth, scientific discovery, and national competitiveness. But without fundamentally transforming the underlying infrastructure, AI's potential will continue to be constrained."
Euclyd's system has yet to be validated in large-scale commercial deployments. Nevertheless, the company claims its silicon system, designed for foundation models, can reduce both energy consumption and costs in AI data center infrastructure, a sector currently experiencing a surge of investment.
The company is planning two main revenue streams: first, selling hardware and physical rack systems to enterprise customers seeking to build secure, on-premises AI inference operations; second, licensing its intellectual property to other companies so they can develop their own chips using the existing technology.
Kastrup noted: "Samsung's contribution to us extends far beyond capital. As one of the world's leading memory chip manufacturers, Samsung brings substantial engineering expertise, deep knowledge of system technology, supply chain familiarity, and an extensive industry network."
Kastrup also disclosed that Euclyd intends to formally launch its physical chip system in 2028, with aspirations to serve thousands of enterprise clients by 2030.